Patients with long-standing inflammatory bowel disease (IBD) have an increased risk of developing colorectal cancer. Performing periodic dysplasia screening and surveillance may diminish this risk. To date, chromoendoscopy is the only technique that has consistently yielded positive results in large, well-designed dysplasia-detection trials. Most major society guidelines endorse chromoendoscopy as an adjunct, accepted, or preferred dysplasia-detection tool. This review outlines the available endoscopic technologies for the detection of dysplasia in IBD, considers the evidence supporting their use, and assesses which modalities are ready for use in clinical practice.
Key points
- •
Patients with long-standing inflammatory bowel disease have an increased risk of developing colorectal cancer. Performing periodic dysplasia screening and surveillance may diminish this risk.
- •
Current surveillance practices, the mainstay of which is white-light examination with targeted and random biopsies, are imperfect, and novel approaches are needed.
- •
Various advanced endoscopic techniques have been studied in an effort to improve the efficacy and efficiency of dysplasia detection.
- •
To date, chromoendoscopy is the only technique that has consistently yielded positive results in large, well-designed dysplasia-detection trials. Most major society guidelines endorse chromoendoscopy as an adjunctive, accepted, or preferred dysplasia-detection tool.
- •
Narrow-band imaging, Fuji Intelligent Chromoendoscopy, i-Scan, autofluorescence imaging, and confocal laser endomicroscopy have yielded conflicting outcomes and are not ready for use in clinical practice.
- •
The widespread use of advanced endoscopic imaging will lead to a paradigm shift in the way gastroenterologists diagnose and treat dysplasia in inflammatory bowel disease.
Introduction
Inflammatory bowel disease (IBD), including Crohn disease (CD) and ulcerative colitis (UC), results from an inappropriate inflammatory immune response to normal intestinal microbiota in a genetically susceptible host. IBD involving the colon predisposes patients to numerous clinical consequences, including an increased risk of developing colorectal cancer (CRC). The precise risk of cancer, which in the past may have been overestimated because of reliance on outdated evidence, remains unclear, with more recent studies estimating the relative risk of CRC in UC to be between 1 and 2.75. Meta-analyses have shown that duration and extent of disease greatly affect the risk of neoplasia, with nearly one-fifth of patients developing cancer after 30 years. Although most of the literature on neoplasia in IBD is based on data from studies of UC, the risk of cancer appears to be similar in Crohn colitis if at least one-third of the colonic mucosa is involved.
The progression to carcinoma in IBD likely stems from chronic inflammation of the colonic mucosa. There are considerable data supporting the hypothesis that carcinogenesis in IBD typically follows a stepwise pattern from inflammation, to dysplasia, to carcinoma. This pattern provides a rationale for screening and surveillance practices aimed at identifying neoplasia at an early stage. Although there have been no randomized controlled trials demonstrating a mortality benefit, there is indirect evidence justifying dysplasia screening and surveillance in IBD. Retrospective studies have demonstrated that colonoscopic surveillance decreases CRC-related mortality in patients with UC. In addition, having undergone 2 or more colonoscopies offers even more protection. Based on such studies, periodic colonoscopic dysplasia surveillance is currently considered the standard of care for all patients with long-standing UC and Crohn colitis.
Although the practice of colonoscopic surveillance is widely accepted, its specific implementation can be controversial. At present the most common surveillance technique involves white-light endoscopic (WLE) examination of the colon with resection or biopsy of any suspicious lesions, as well as random 4-quadrant biopsies taken every 10 cm throughout the length of the colon. The rationale for obtaining nontargeted biopsies is based on the observation that dysplasia in IBD can be difficult, or impossible, to detect using standard endoscopic equipment. To achieve adequate sensitivity for dysplasia detection in flat colonic mucosa, it has been estimated that between 33 and 64 random-biopsy specimens must be obtained at colonoscopy.
However, it is important to point out that most data supporting the random 4-quadrant biopsy methods predate modern endoscopic equipment, and numerous recent studies have called this technique into question for several reasons. First, obtaining the requisite 33 random biopsies still evaluates only a small fraction of the colonic mucosa. Second, practicing gastroenterologists may not strictly follow the random-biopsy protocol. A survey of more than 300 gastroenterologists in the United States found that nearly half of the respondents took less than the recommended number of biopsies, which may stem from the significant time and cost associated with the protocol. Third, the random-biopsy protocol has repeatedly proved to be a very low yield technique. Van den Broek and colleagues reported a dysplasia-detection rate of 0.2% for random biopsies compared with 23% for targeted biopsies. Moreover, only 1 of 475 (0.002%) patients had a change in management based on the results of a nontargeted biopsy. Fourth, in accordance with the findings of van den Broek, investigators are finding that most dysplastic lesions in IBD are in fact detectable using modern endoscopic modalities. Finally, there is evidence that the rates of CRC and dysplasia may be decreasing in UC patients overall. This smaller, more elusive target may further limit our ability to detect existing neoplasia using outdated techniques.
Frustration with current surveillance practices and a sense that the dysplasia target has shrunk have led investigators to seek more innovative ways of approaching this problem. Our growing understanding of the natural history of dysplasia and CRC in IBD has paralleled the evolving sophistication and resolution of endoscopic tools used to detect it. Naturally the two have become intertwined, and novel endoscopic modalities have been used in the quest to uncover neoplasia. This review aims to outline the available endoscopic technologies for dysplasia detection in IBD, consider the evidence supporting their use, and assess which modalities are ready for use in clinical practice.
Introduction
Inflammatory bowel disease (IBD), including Crohn disease (CD) and ulcerative colitis (UC), results from an inappropriate inflammatory immune response to normal intestinal microbiota in a genetically susceptible host. IBD involving the colon predisposes patients to numerous clinical consequences, including an increased risk of developing colorectal cancer (CRC). The precise risk of cancer, which in the past may have been overestimated because of reliance on outdated evidence, remains unclear, with more recent studies estimating the relative risk of CRC in UC to be between 1 and 2.75. Meta-analyses have shown that duration and extent of disease greatly affect the risk of neoplasia, with nearly one-fifth of patients developing cancer after 30 years. Although most of the literature on neoplasia in IBD is based on data from studies of UC, the risk of cancer appears to be similar in Crohn colitis if at least one-third of the colonic mucosa is involved.
The progression to carcinoma in IBD likely stems from chronic inflammation of the colonic mucosa. There are considerable data supporting the hypothesis that carcinogenesis in IBD typically follows a stepwise pattern from inflammation, to dysplasia, to carcinoma. This pattern provides a rationale for screening and surveillance practices aimed at identifying neoplasia at an early stage. Although there have been no randomized controlled trials demonstrating a mortality benefit, there is indirect evidence justifying dysplasia screening and surveillance in IBD. Retrospective studies have demonstrated that colonoscopic surveillance decreases CRC-related mortality in patients with UC. In addition, having undergone 2 or more colonoscopies offers even more protection. Based on such studies, periodic colonoscopic dysplasia surveillance is currently considered the standard of care for all patients with long-standing UC and Crohn colitis.
Although the practice of colonoscopic surveillance is widely accepted, its specific implementation can be controversial. At present the most common surveillance technique involves white-light endoscopic (WLE) examination of the colon with resection or biopsy of any suspicious lesions, as well as random 4-quadrant biopsies taken every 10 cm throughout the length of the colon. The rationale for obtaining nontargeted biopsies is based on the observation that dysplasia in IBD can be difficult, or impossible, to detect using standard endoscopic equipment. To achieve adequate sensitivity for dysplasia detection in flat colonic mucosa, it has been estimated that between 33 and 64 random-biopsy specimens must be obtained at colonoscopy.
However, it is important to point out that most data supporting the random 4-quadrant biopsy methods predate modern endoscopic equipment, and numerous recent studies have called this technique into question for several reasons. First, obtaining the requisite 33 random biopsies still evaluates only a small fraction of the colonic mucosa. Second, practicing gastroenterologists may not strictly follow the random-biopsy protocol. A survey of more than 300 gastroenterologists in the United States found that nearly half of the respondents took less than the recommended number of biopsies, which may stem from the significant time and cost associated with the protocol. Third, the random-biopsy protocol has repeatedly proved to be a very low yield technique. Van den Broek and colleagues reported a dysplasia-detection rate of 0.2% for random biopsies compared with 23% for targeted biopsies. Moreover, only 1 of 475 (0.002%) patients had a change in management based on the results of a nontargeted biopsy. Fourth, in accordance with the findings of van den Broek, investigators are finding that most dysplastic lesions in IBD are in fact detectable using modern endoscopic modalities. Finally, there is evidence that the rates of CRC and dysplasia may be decreasing in UC patients overall. This smaller, more elusive target may further limit our ability to detect existing neoplasia using outdated techniques.
Frustration with current surveillance practices and a sense that the dysplasia target has shrunk have led investigators to seek more innovative ways of approaching this problem. Our growing understanding of the natural history of dysplasia and CRC in IBD has paralleled the evolving sophistication and resolution of endoscopic tools used to detect it. Naturally the two have become intertwined, and novel endoscopic modalities have been used in the quest to uncover neoplasia. This review aims to outline the available endoscopic technologies for dysplasia detection in IBD, consider the evidence supporting their use, and assess which modalities are ready for use in clinical practice.
Dysplasia-detection techniques
White-Light and High-Definition White-Light Endoscopy
Two large retrospective studies have demonstrated that most dysplasia in IBD is visible using standard WLE. In their institutional experience, Rutter and colleagues reported that 77.3% of neoplastic lesions were macroscopically visible at colonoscopy. On a per-patient basis, 89.3% of patients had macroscopically detectable neoplasia. Rubin and colleagues achieved similar results at their institution, reporting per-patient sensitivities for dysplasia and cancer of 71.8% and 100%, respectively. Based on these studies, as well as others reporting similar results, it is currently accepted that the majority of neoplastic lesions in UC patients are visible with modern endoscopic equipment. This paradigm again highlights the questionable value of obtaining nontargeted biopsies of endoscopically normal-appearing colonic mucosa.
The advent of modern high-definition (HD), or high-resolution, colonoscopy has further improved the quality of white-light examinations. HD refers to the pixel density of a system. Standard-definition endoscopes produce an image signal of 100,000 to 400,000 pixels, whereas high-resolution or HD endoscopes produce signal images with resolutions that range from 850,000 pixels to more than 1 million pixels. Resolution is distinct from magnification, which involves simply enlarging an image without affecting pixel density.
HD colonoscopy has been shown to improve adenoma detection rates in average-risk patients by improving the ability to detect subtle mucosal changes. Subramanian and colleagues compared the yield of dysplastic lesions detected by standard WLE with that of HD endoscopy in a retrospective cohort of patients with UC or Crohn colitis. An adjusted prevalence ratio of detecting a dysplastic lesion on targeted biopsy of 2.99 (95% confidence interval [CI] 1.16–7.79) for HD colonoscopy was found. WLE is likely to continue to play a vital role in dysplasia detection in IBD with the continued development of HD colonoscopy technologies ( Fig. 1 ).
Most dysplasia in IBD is visible using modern WLE equipment. HD endoscopes will continue to make WLE an important tool for the detection of dysplasia in IBD.
Narrow-Band Imaging
Narrow-band imaging (NBI) is a technology that uses specialized light filters to modulate the intensity of various constituents of the white-light spectrum. Most NBI systems have narrow-band filters with the absorption property of hemoglobin that allow them to highlight vascular architecture and structures of the mucosal surface. NBI is widely available on newer model endoscopes, making it an attractive option as an adjunct to current dysplasia-detection tools in IBD. Unfortunately, most well-designed randomized studies have failed to show a difference in dysplasia detection using NBI in comparison with WLE.
Dekker and colleagues performed a prospective, randomized, crossover study in which 42 patients with long-standing UC underwent colonoscopies with both NBI and WLE ( Fig. 2 ). Neoplasia was detected in 4 patients by both techniques, in another 4 patients by NBI only, and in another 3 patients by WLE only. These results showed no statistical difference between the two modalities. Of note, significantly more nonneoplastic, or false-positive, biopsies were taken using NBI. A similarly designed study compared a newer-generation NBI system with HD WLE in patients with UC. Again, there was no difference between the two modalities in the number of dysplastic lesions detected. A more recent parallel-group trial of patients with UC randomized to receive colonoscopy with NBI or WLE was discontinued early because a prespecified midpoint analysis showed no difference in the primary outcome of dysplasia detection between the two groups.
The dysplasia-detection capability of NBI has also been compared with that of chromoendoscopy. In a prospective, randomized, crossover study, Pellisé and colleagues enrolled 60 patients with UC to undergo colonoscopy with both chromoendoscopy and NBI in random order. The study showed a similar dysplasia-detection rate for NBI and chromoendoscopy. However, NBI also resulted in a higher rate of missed lesions, with an odds ratio for missed neoplasia of 4.21. Although this did not reach statistical significance, it was a key finding. Based on the high miss rate (ie, high specificity but low sensitivity) and because the study was somewhat underpowered, the investigators concluded that NBI cannot be recommended for use in dysplasia surveillance in IBD patients at present.
There are likely several reasons for the shortcomings of NBI in the realm of dysplasia detection in the IBD population. According to Pellisé and colleagues, the lack of sensitivity may be attributed to the fact that NBI principally evaluates vascular patterns rather than alterations in crypt architecture detected by chromoendoscopy. The disruption of the vascular pattern is a well-recognized consequence of chronic colitis, and may hamper the efficacy of this technology in this setting. The importance of vascular-pattern analysis in dysplasia detection is not entirely clear, and analysis of crypt architecture and pit patterns has been more readily accepted in identifying neoplastic lesions. A simpler explanation is that NBI provides a darker image than WLE, which may actually hinder visualization of mucosal lesions.
NBI is a widely available and convenient endoscopic imaging modality. However, several studies have failed to demonstrate its efficacy for the detection of dysplasia in IBD patients.
Chromoendoscopy
Chromoendoscopy is the topical application of dyes to the colonic mucosa to enhance detection and delineation of surface abnormalities. The two dye agents most commonly used in trials, as well as in clinical practice, are methylene blue and indigo carmine. Methylene blue is adsorbed and absorbed by normal colonocytes but not by inflamed or neoplastic colonic mucosa, thus highlighting the pit patterns of mucosal lesions. Indigo carmine is not absorbed by the mucosa but simply pools within colonic crypts, allowing for the demarcation of inflamed or neoplastic lesions. These agents are applied to the entire colonic mucosa using a spray catheter or the water jet channel on a standard colonoscope, and allow for inspection within minutes of application.
Chromoendoscopy has been comprehensively studied as a “red-flag” technology for dysplasia detection in IBD. In a prospective cohort of patients with pancolitis, Matsumoto and colleagues showed that chromoendoscopy using indigo carmine significantly enhances the detection and characterization of flat lesions harboring dysplasia. Kiesslich and colleagues performed a randomized controlled trial comparing conventional colonoscopy with chromoendoscopy using methylene blue in 165 patients with long-standing UC. Chromoendoscopy led to a 3-fold increase in the detection of intraepithelial neoplasia in comparison with conventional colonoscopy (32 vs 10; P = .003), and allowed for the differentiation between nonneoplastic and neoplastic lesions with sensitivity and specificity of 93%. The same group of investigators performed a follow-up study in which UC patients were randomized to conventional colonoscopy or chromoendoscopy with endomicroscopy. Chromoendoscopy led to the detection of 4.75-fold more neoplasias ( P = .005) than conventional colonoscopy. Endomicroscopy further enhanced the ability to accurately characterize lesions as neoplastic (accuracy 97.8%). Hurlstone and colleagues compared magnification chromoendoscopy in 350 patients with UC with a group of matched historical controls who had undergone conventional examinations. Significantly more neoplastic lesions were detected in the chromoendoscopy group than in the controls (69 vs 24, P <.0001). In a tandem study, Rutter and colleagues performed “back-to-back” white-light and indigo carmine colonoscopic examinations in a cohort of 100 patients with UC, and found 7 dysplastic lesions with chromoendoscopy that were missed during the WLE examination. A similarly designed tandem colonoscopy study by Marion and colleagues demonstrated that chromoendoscopy using methylene blue was able to detect more dysplastic lesions than either standard WLE targeting or random biopsies. Targeted biopsies with dye spray yielded 17 dysplastic lesions, compared with 3 lesions from random biopsies and 9 lesions with targeted WLE biopsies ( Fig. 3 ).
